We have made a number of measurements, impossible at all but a handful of sites in the world, in which the electrical and fMRI signals are assessed simultaneously. We have done this in awake behaving primates implanted with various types of electrodes. In one study, we used linear arrays of electrodes implanted into the cerebral cortex, including the primary visual cortex. In that study, we were able to measure the laminar distribution of neural activity and relate it to the BOLD signal measured in and around the electrode. We found that the functional coupling was not consistently predicted by any particular cortical layer, and that the measured neural responses were more consistent across trials than the BOLD response. These observations probably reflect the fact that the BOLD signal is susceptible to large-scale, global fluctuations spanning nearly the entire cerebral cortex. These slow changes are superimposed on the local response, adding considerable variance to the measured fMRI signals. The global signal does have a neural correlate in one particular frequency range of the local field potential. We have recently written a comprehensive review on our studies related to this topic, published this year. We are also investigating the relationship between neural spiking and the BOLD signal under different perceptual conditions. In a previous study, we showed that during perceptual suppression, where a visual stimulus is physically present but temporarily escapes perception, fMRI signals and electrophysiological signals in the primary visual cortex become uncharacteristically divergent. Specifically, the fMRI signal reflected the monkeys perceptual state, whereas the spiking of individual neurons in the same region of the brain reflected the presence of the physical stimulus, but did not change their activity according to the reported perception. This raises the question: which aspects of spiking correlate with the fMRI signal and which do not? This question is particularly salient in the inferotemporal cortex of the monkey, where neurons respond to complex object features. We have recently begun to record signals using implanted MR-compatible microwire bundles. These microwires are able to isolate and hold single neurons for extended periods of time while we measure fMRI signals. One property of the neurons we have recorded is that they have very different response properties. For example, when we show the monkeys complex videos, adjacent neurons fire at very different points in time. How would the BOLD signal correspond to the firing of any one particular neuron, of all the neurons respond differently? Would it reflect the mean firing over a population, or are some neurons more critically related to the BOLD responses than others. These and several similar questions can be addressed by simultaneously recording spiking and fMRI responses from the inferotemporal cortex in monkeys presented with complex stimuli. Finally, in a recent study, we demonstrated that not only is the relationship between neural activity and fMRI fluctuations difficult to characterize, but it is also unstable over time. That is, over periods of seconds and minutes, the functional coupling between the electrical and blood-based signals changes. We established this using implanted electrocorticogram (ECoG arrays) measuring electrical signals inside the MR scanner. We found that the correlation level between the two signal types waxed and waned over time. This observation is an important clue as to the basis of functional connectivity dynamics, in which the level of voxel-to-voxel correlation across brain regions changes over time.